Fall Detection Algorithm using Body Angle for Accurate Classification of Falls and ADLs

Falls in elderly are an important issue for the aging populations around the world. They cause severe physical and emotional injuries and surge healthcare and hospitalization costs. Detection of falls in time help reduce the consequences of the fall by providing a faster rescue to the patient which prevents more serious injuries. The present research describes a threshold-based algorithm to distinguish between falls and Activities of Daily Living (ADL). This algorithm was implemented on a wearable device attached to the waist. Three thresholds are used in order to distinguish a fall from any other activity. Falls are simulated by 10 young volunteers under supervised conditions and ADLs are performed by 10 elderly subjects. Experimental results show that the system detects falls compared to ADL with a sensitivity and specificity of 93.3% and 100% respectively.